In this post, I challenge the financial image of HHAs presented to congress by MedPAC.
In my last post, we explored the efforts by home health advocates to convince congress to eliminate cuts to HHAs through bills presented in the congress and senate referred to as “Preserving Access to Home Health”.
In this previous post, we talked about efforts through NAHC and other industry advocates to convince congress to eliminate these cuts as they threaten the survival of many agencies. This is done through efforts by these advocates to get home health stakeholders to contact their senators and congressmen and tell them how they are impacted and why this issue is important.
On the other side of the table, there is MedPAC, an entity whose purpose is to research healthcare spending by the federal government and report their findings to congress.
These reports show, through data collected and published by MedPAC, that home health is by far the most profitable provider group supported by CMS under the Medicare program. Not only do they support these cuts, but they believe CMS should be cutting home health spending even more.
MedPAC bases nearly all of this analysis on data found in cost reports. This is the same data I have been using in my blog since October of last year (2023). MedPAC produces an annual report for congress each year they refer to as the “Data Book”. This report covers all provider types and includes reporting on Medicare Advantage and other healthcare related programs administered by HHS.
The most recent comprehensive report was published last July, 2023. Chapter 8 focuses on home health and can be downloaded as a separate document. My interpretation of this report is that MedPAC believes that in general, HHAs are overpaid and inefficient. Not only do they support the permanent adjustment cuts by CMS, they support additional cuts intended to address the liability associated with the CMS definition of budget neutrality (Temporary Adjustment). Here is the first thing you see when you read this chapter:
This image stands alone at the beginning of the report. It is as if they were trying to share their most important point with legislators too lazy to read the report beyond the first page of the chapter.
This recommendation is largely based on the data on home health margins presented by MedPAC in this chart. These margins are calculated through the same cost report data I am using and from the same source:
A few days ago, I updated my Sisense BI environment with the latest HCRIS data from the cost reports, the same data that MedPAC is looking at. These results are basically the same as I have presented in previous posts on this subject. This is what I get for margins, including 2022, which was not available yet to MedPAC when their report was written:
Not only do I come up with significantly lower margins, but the cost report data shows these margins are declining while MedPAC shows them growing each year using the same data. For 2021, they show 24.9% and I come up with 13.10%, nearly half of the MedPAC margin.
When I first saw this disparity, a couple years ago, I spent considerable time reexamining my own work.
Profit margins are not difficult to calculate.
Profit margin = (revenue - expenses) / revenue
Basically you just need two values, revenue and expenses. As simple as this seems, it can get a little more complicated when you have to make decisions on where this data is coming from in the cost report and which cost reports should be included in the results.
In the charts that I have provided in my posts using home health cost report data, I have intended to use standards for processing and characterizing this data. When it comes to the values used to represent general financial terms, it is important to define them and where they come from.
In this post, I will show you step by step how I came to these profit margins using the same cost report data used by MedPAC. If anyone wants to test this process, you can use it to identify the data in your own cost reports and see if you feel it appropriately represents your own Medicare margins. If it does, and I have used the same process to calculate everyone’s combined margins the same way, then why are my results so different from MedPAC’s?
MedPAC has a well educated and experienced group of commission members and staff. Their budget request for 2024 was just over $13 million. I would think that if one guy sitting in a home office can put this effort together, they should be willing to provide the same level of detail in support of their data the next time they publish it. There is too much at stake to not make an effort to get a true understanding of what is going on in home health. This starts with transparency when it comes to presenting charts like these. I will take my turn in clarifying my results now.
Revenue
The home health cost reports are structured like tax returns. They include the individual line items for revenue and expenses that are also represented as totals. Home health cost reports have total revenue and individual totals for Medicare, Medicaid and Other. In the chart represented by MedPAC and my chart we are using only Medicare revenue. This can be found on worksheet F-1 Statement of Revenue and Expenses, along with Operating Expenses.
Expenses
The Medicare revenue for each annual cost report comes straight from this value in the data, worksheet F1, row 1, column 1.
For expenses, we need to calculate Medicare expenses only. All we have in the cost report is total operating expenses. We can’t use the ratio of revenue for the different pay classes and apply them to the expenses because we all know that revenue by financial class is not equal for the same services.
Fortunately, the cost reports include statistics as well as financial data. They include census also broken down by Medicare, Medicaid and Other. Assuming that census means the same thing in all three settings, we can come up with a revenue and operating expense per census.
The census statistics come from worksheet S-3 Statistical Data. The cost reports itemize these by the clinician disciplines, but we are just using Medicare total census for the year.
Margins
Dividing revenue and expenses by the census gives us a method of comparing Medicare only revenue to Medicare only expenses. It assumes that the same level of services (costs) are provided for each financial class per census. Although this is not entirely true, the differences can’t account for any significant differences in margins at the census level.
This is how I came up with a profit margin for my report chart. It works for your own cost report and all of them put together. If any of you feel that this is incorrect or if you feel there is a better way of calculating Medicare profit margin, please describe it to me and we can do it differently.
Once these terms and the associated calculations are out in the open, anyone can reproduce the same results using cost report data. If two sources provide different answers, the only explanation could be that different data was used or the calculation was different. If this is the case, this will be evident when the supporting data is provided.
Discrepancies
As I mentioned, I spent a considerable amount of time trying to figure out what could explain these dramatically different pictures of home health profitability. For the sake of argument, let’s assume I made some sort of error that produced these results.
In order to come up with the lower margins, I would have to have used a lower revenue figure than MedPAC or a higher expense total.
The revenue is straightforward and comes directly as a total from the cost report. MedPAC mentions that their margins include only regular revenue, reimbursement from emergency pandemic funding is not included. If I had added it by mistake, my margins would have been higher. I made no adjustments to the revenue provided as total Medicare revenue, so I can’t see how I could have excluded revenue included by MedPAC.
As far as expenses, I used only operating expenses. This does not include depreciation and other expenses that might appear on your own financials. Medicare refers to the expenses as “allowed expenses” in that you can only include the specific items listed in the cost report. For example, “entertainment” is not included although it might appear on your own financials as a legitimate expense.
Again, I have not found any way I could have included expenses not considered by MedPAC.
The other possibility is the process of cleaning the data and what data is used for your calculations. The cost report data is far from completely clean and correct. I have maintained a long list of questions and observations on this subject. In a previous post, “Cleaning Cost Report Data”, I described in detail the steps I take to remove the cost reports I consider to be invalid. These steps can be repeated by anyone.
Here is the criteria I used to exclude cost reports I used for my calculations.
Exclude any reports that are header records only without any numeric data elements
Exclude any reports without clinician costs
Exclude any reports with a total census or visits of zero
For 2021, these rules take the original file of cost reports from 10,361 records to 6,568. In Sisense, these records are not removed or altered, they are suppressed in the calculations. This allows me to reverse or revise these rules. I can also provide a list of the original data, available to everyone, and a list of the records suppressed by each rule in case the rules are questioned.
This is in contrast to MedPAC and CMS who provide only vague references on how they accomplished this. They obviously could not use reports either that did not contain any numeric data, how did they handle this?
CMS briefly discusses data cleaning in the proposed rules, but this is not addressed in the MedPAC reports. CMS describes “eliminating outliers of the top and bottom 5%” without describing what the 5% metric applies to. They also describe “using zeros when no numbers are present in the report”. This is not how you prepare data for analysis.
Let’s say you want to apply business analytics to your EMR data. You might create similar rules to eliminate records like those prior to a specific date when you converted systems or a significant change occurred in the data collected. You would apply these rules to the process that extracted data from your EMR and then you would use the results. You would not simply cut out records based on the extremes of some particular metric. Whatever method you chose, even if you were just using the data internally, you would include a description of what data was included and what was not when presenting your findings to others.
Unlike your EMRs, we are not talking about thousands and maybe millions of records. In my results, I have used 6,568 cost reports for 2021 and excluded the remaining 3,793 based on three rules. I can document this and it can be repeated by others. This is the scientific approach that should be taken with data used for such an important purpose. If you are going to publish data and make decisions involving billions of dollars in connection with this data, the process and calculations you used should be clear and available to all interested parties.
Despite the fact that the cleaning process used by MedPAC and CMS, and the results of it, remain a mystery, it is unlikely that this process could explain the large variance in our results. It has to be something else. As it stands, I have documented how I have come up with my margins, it is MedPACs turn to describe how they came up with theirs.
In my next post, we will continue to explore issues with the MedPAC report as it applies to home health. As important as it is that these margins represent reality, it is even more important that the image of financial health for HHAs include all aspects of their business, not just Medicare. MedPAC includes “all payer” margins when it comes to hospitals, for home health, it is missing. We will explore the impact of this omission on the home health data provided to congress and why it might be happening.
Comentários